Food distribution in Mandera district (Kenya)

By Manuel Duce-Marques

The author, Manuel Duce-Marques is a qualified nurse who spent several years working for MSF Spain in Mandera District, Northeast Kenya. He has just recently graduated with an MSc in Human Nutrition from the London School of Hygiene and Tropical Medicine.

Mandera is one of three districts in the North-eastern Province of Kenya. It is an area prone to drought and food insecurity and has a population of 131,000 with 37,900 estimated as living in Central Mandera. The district is geographically isolated and has weak links with other districts. Its people are ethnic Somalis who are traditionally nomadic pastoralists but now fall more into the category of agro-pastoralists as they practice some cultivation of staple crops - maize and beans. Central Mandera has been a settlement for Somali refugees since 1991 and about 10,000 of these refugees still live in the former refugee camps which are now considered official locations. During the severe drought of 1991-2 livestock holdings were drastically reduced so that many families became destitute and were forced to migrate to Central Mandera.

Between 1994-6 there were three successive rain failures in the district. However, despite a worsening 'alert' status^* in the district, indicated by the Arid Lands Resource Management early warning system and a series of MSF nutritional surveys, the Kenyan government was reluctant to declare an emergency. MSF Spain, who were initially contacted by a local NGO in March 1996 because of its concerns about the plight of pastoralists in the district, attempted to convince the Kenyan government, UN agencies (WFP) and other NGOs in Nairobi to mount an emergency food intervention. This advocacy did not prove effective. In May 1996, MSF recorded a 32.4% prevalence of malnutrition in Mandera. An emergency was not officially declared until February '97, and therefore no request for emergency food aid was made until this date. In response to the deteriorating situation in Mandera, MSF took the decision to step outside its usual mandate and implement a general food distribution programme. An MSF team already present in the area was given the task of designing and implementing the programme although they had no prior experience of this type of intervention. It was hoped that this would encourage other organisations, with experience of general food distribution, to take over the programme.

Registration

The main objective of carrying out a registration was to obtain 'reliable' figures for planning the general ration distribution. Beneficiary numbers obtained from other sources were often contradictory and believed to be inflated. Two different types of registration procedure were selected for Fino division and Central Mandera.

Central Mandera

The population in Central Mandera (about 37,900 people) consisted of displaced persons and refugees/returnees. It was clear that not everybody was equally affected by the drought so that there were different levels of food insecurity. It was therefore necessary to establish criteria for who should or should not be registered for the general ration distribution.

After some consideration, it was decided to target ration cards to families with a malnourished child in one of the feeding centres (although it was recognised that this may not necessarily target the most food insecure families as malnutrition may be related to disease or poor caring practices). This led to an immediate increase in admissions to the feeding centres. While this could be construed as an advantage as feeding centre coverage improved, in part the increase was due to cheating by registering the same child more than once and by staff succumbing to community pressures to admit children who did not fit the criteria. It also created negative consequences within the centres, e.g. overcrowding resulting in increased risk of cross infection, reduced staff / beneficiary ratio, reduced quality of supervision and increased levels of dissatisfaction among mothers, leading to an increased proportion of defaulters. As the situation ran out of control, the distribution of ration cards was stopped. MSF could not find a way around the problem of over-registration and ration cards were not distributed again. The total number of ration cards which had been distributed were about 3,000, which corresponded to a population coverage of 18,000.

Fino division

In Fino division it was decided to implement a full registration (once off) for the entire population. To reduce the potential for multiple registration, the exercise was carried out simultaneously on one day for the 7 villages close to Somalia and on the following day for the 8 villages close to Ethiopia. The total number of ration cards distributed was 7,698 which was equivalent to 26,900 beneficiaries.

Registration Process

Resources

After planning a rough schedule for the registration and obtaining agreement with the community, the programme had to be postponed for two months until the resources became available. All the telecommunications material (radios) needed for the security of the teams 'disappeared' in the Nairobi customs. This, plus the bureaucratic red-tape that had to be negotiated to get all the material back, added to the anxiety and impatience amongst locals and increased pressure on our (MSF) teams as well as their own frustration at the delays.

Timing

As some of the registration points were close to Somalia and Ethiopia, it was decided to inform leaders of the communities only a few days in advance about the registration day to avoid artificial movements of population from neighbouring countries and within Kenya itself. It was recognised that this might have prevented some legitimate groups from registering but was still considered necessary to ensure effective targeting of resources. However, in spite of this precaution, the two month's delay allowed small new villages to spring up around locations which were scheduled to be included in the registration.

System used

Families were made to sit in lines and wait some distance away from the registration point. Community leaders would send a few families at a time to the clerks to avoid overcrowding. During the registration process, every member of the family put one finger in highly concentrated Gentian Violet (GV) to indicate ration card receipt thereby avoiding double registration. However, the main problem was that a very similar method was used in the elections in Kenya and people knew very well how to clean off the dye. Before registration they greased their fingers with coconut oil and afterwards they cleaned the GV with fuel and lemon. It was obvious that the plan had to be changed. Chiefs were therefore asked to gather all the population in one place early in the morning where each member of the family would be given a piece of paper with an MSF stamp on it (this was difficult to duplicate). The paper would then be collected during registration. This proved to be a far more effective method than the previous one.

Registration form

Ration cards serve the dual purpose of ensuring that the beneficiary receives the food and also facilitates control over the distribution.

The information on the ration card was rewritten in a register: card number, name of the head of family, number of dependants, number of children under five and number of pregnant and lactating women. MUAC measurements of all the children under 5 were initially undertaken but this led to crowding and confusion, so the practice was stopped.

Security

Registration of large numbers of people at one point can take a long time. As the day wears on discomfort for beneficiaries and programme implementors increases. Therefore minimum conditions of security are needed to ensure success. Responsibility for security was given to the chief of each location. They were given information about how the registration would be carried out and the roles of the various people involved. In addition, the community of each location was responsible for ensuring that the registration point was kept 'intact'. They would often prepare fences around the site made with bushes. In the few locations where different clans were represented there would often be a dominant group who influenced the proceedings, trying to register favoured persons more than once. This would lead to mounting tensions (in rare cases leading to violence) and sometimes we could not finish the registration which probably meant that the most vulnerable families (who were likely to be last in the queues) did not receive the ration cards.

Conclusion

• The registration process was efficiently carried out with the help of the local authorities and traditional leaders and there were only a few security problems. However, we found that dishonesty amongst the staff and organised pressure from powerful people meant that even this method of registration could only limit rather than entirely prevent abuse .
• The cost of the registration was not high as most of the logistical infrastructure was already in place in the programme and extra-expatriates were seconded from other MSF programmes in the area. However, the use of these staff meant that some feeding centre activities had to be curtailed.

General Ration Distribution

Two members of MSF came especially to take charge of the food distribution in Fino Division. However everybody in the team was involved; the field co-ordinator was in charge of all negotiations with local government and other players, the MSF administrator was in charge of buying food items and renting all the cars and lorries, the logistician for the feeding centres was in charge of the storage of the food. Distribution was directed to the family unit. A family member collected the ration at a distribution point. The family ration was measured by volume by MSF staff using scoops after presentation and verification of the ration card. The card was stamped and marked for each item collected by making a hole in a specific place. The ration consisted of 4.5 kg of maize, 3 kg of beans, and 1 kg of oil per person per month, which was equivalent to 1100 Kcal/day. In addition to these basic commodities, 7.5 Kg of a blended food (UNIMIX), was included for children under five years of age and pregnant and lactating women. This contributed a further 1000 Kcal/day. The ration was set at this level based on the theoretical quantity required and what was feasible at field level, given resource and logistical constraints. A local teacher from each distribution point was trained to teach mothers about the correct preparation of the UNIMIX.

Central Mandera

The first distribution was carried out at the end of November. The distribution points were the supplementary feeding centres. Food distributions were carried out once a month. To prevent families trying to collect the food at more than one feeding centre, the distribution was carried out in all three centres at the same time. In view of the fact that it would be easy to exaggerate numbers in a family a standard ration card for 6 members was distributed to beneficiaries. Feeding centre staff did the distribution. However as they were part of the community they were prone to giving more food to members of their own tribe. Plans were therefore made to move staff from one centre
to another, but as it happened this was not necessary as our capacity to control staff activities improved.

Fino Division

In most cases there was one distribution point for each location which minimised both the number of people who attended any distribution at one time and the distance travelled to load the food. Because of the large number of distribution points and the logistic capacity the distribution was carried out monthly. The food needed for Fino division for one month was about 350 tons, ranging from 2 to 35 tons for the different locations.

Distribution Procedure

Staff

The distribution staff consisted of the 2 expatriates, 20 local MSF staff and between 4-6 watchmen. There were also loaders from the local area.

Distribution point

The recipient community waited some distance away from the distribution point to minimise overcrowding around the distribution point. The structure of the distribution centre was made from wood and was located close to some existing local infrastructure (e.g. school, etc.). On the day of the distribution a plastic sheet was placed on the roof to protect the team from the sun. The simple structure allowed for crowd control and there was one entrance and one exit. Two lines of people were served in parallel. Each line required one clerk to check the cards with the register books; 1-2 people to distribute each commodity and 1 supervisor. One expatriate staff member remained with the clerk with another at the end of the chain.

Food losses

A small percentage of losses (around 10%) was considered acceptable. However, commodities of high value (e.g. oil), were more likely to disappear. Food losses occurred for a variety of reasons:

• Losses from the lorry: we counted the numbers of bags in the lorries before and after the trip. If some bags were lost during the trip the value was discounted from the rent of the trucks or from the driver's salary. However, a frequent practice was to make holes in the bags, and then collect the food from the floor of the lorry. This could result in hundreds of kilos of grain being lost.
• Diversion by staff and chiefs: the staff were given 3 warnings if they were suspected or seen to be mismanaging the food in the distribution. After the third warning they would be sacked without compensation. However replacement staff were often caught doing exactly the same as their predecessors.
• Over-Scooping: scoopers might give fuller scoops to relatives or friends. The accumulated loss meant that many received less than they should have. The only way to control this was through randomly weighing some of the bags received by the family representative and comparing the weight with the number of people who were supposed to receive the food. Surplus was removed and low weight bags were replenished. This task was very labour-intensive for both the scoopers and the controllers.
• Altering Ration Cards: owners of a ration card might increase the number of beneficiaries on their cards (e.g. if there were 5 on their card they wrote the number one in the front to make it appear as if it were 15). So, careful control of each card was necessary. We found that in some cases our own staff changed the number in the book.
• Packing errors: sometimes low weight bags were sent from the factory. This was attributed either to water losses from the maize (when it is packed fresh and subsequently dries out) or to the bag being under-filled in the factory.
• Poor quality foods: other losses occurred when some food items (mainly UNIMIX) had to be discarded due to poor quality or spoilage. Poor quality might be due to poor quality food being packed, poor transport conditions and lack of adequate ventilation systems in the stores. MSF managed to improve storage conditions.

Security

As there is no physical or controlled border with Somalia, Mandera town and the border areas are considered to be unsafe. Bandits regularly make raids on Kenyan territory. The risk of attack by Somali groups during the distribution was therefore very high. MSF policy on security was very clear, i.e. not paying for security and not enlisting armed guards for MSF vehicles (with the exception of Somalia). Insecurity presented a number of problems for the programme.

1. Security during the distribution was the responsibility of the local chief/traditional authority of each location. Their duty was to negotiate with local government to obtain enough security to guarantee a safe distribution. MSF stressed that at the first sign of insecurity, distribution would be stopped. In most cases the distribution system worked very well, but in a few locations after the relatives of the local leaders had received their rations, their interest in the efficient functioning of the distribution system waned and powerful individuals were able to take more food than they were entitled to leading to beneficiary dissatisfaction and unrest. MSF were then forced to withdraw. MSF´s policy of abandoning an area under these conditions, sometimes led to undermining the power of some corrupt leaders. MSF would return the following month to try again and the community would attempt to change the leadership to ensure more secure future distributions.
2. MSF decided to start without any protection for the food convoys. However, following a number of attacks by bandits coming from Somalia on vehicles, the decision was taken to stop the distribution of food until local government could provide, at their own costs, the security necessary to guarantee us the distribution. This was explained to the leaders at all the locations who also informed local government. This was a shock for the government as all other organisations in the area (including UN agencies) paid for this service. After three weeks of discussions (and no distribution) MSF and local government arrived at an agreement. Local government paid for police protection and each of the local ministries in Mandera agreed to provide cars with petrol. However, these measures created their own problems e.g. often the car didn't have petrol, difficulties in finding the person to authorise the payment for the petrol, cars arriving late etc. Due to the high level of insecurity in the area and frequency of theft, we were advised the cars had to be rented in the same location. These cars were in a bad condition and needed constant attention.

Other problems

• Lost ration cards: an individual reporting a lost ration card at the distribution had to wait until the distribution was finished. If another potential beneficiary turned up with the reported lost card, he/she was also told to wait until the distribution was finished. The decision about the rightful owner of the card was made together with the village leader. If nobody turned up with the lost card, a new card was prepared (with a special mark to ensure the recipient could not claim ownership of two cards at the next distribution).
• How to check who is pregnant? What is the limit for lactating women? Pregnancy had to be visually confirmed, and a mother with a child of about 1 year would be classified as a lactating woman. However, the percentage of pregnant and lactating women was clearly over-reported. This was a problem that was very difficult to overcome.
• The quality of the distributed beans was variable, which caused complaints amongst some beneficiaries and on occasions created delays during the distribution.

Main Findings

Our experience showed that a certain level of food losses was inevitable and that powerful, well connected households often received more than their fair share of the ration, usually at the expense of the weak and vulnerable. Security and monitoring mechanisms could reduce some of these abuses but we were forced to be realistic about the extent to which we could effect change. Some leaders forced households to pay to them a part of the ration received. We did not attempt to interfere with this practice. We learned many lessons about how to improve the efficacy and equity of the food distribution system, but came to realise that there were certain aspects of the distribution chain over which we had only limited control. While the objective of providing the general ration was to ensure nutritional well-being, it was clear that a number of households sold the food and therefore used the food as an income transfer, which allowed expenditure on non-food items. This created disappointment amongst some expatriate staff who assumed that the beneficiaries would eat all of what was provided. While, from my point of view, this exchange of rations can be acceptable and sometimes necessary, the process can also be inefficient as the money/goods exchanged for the food is less than the monetary equivalent of its nutritional value.

The two expatriates in charge of the organisation of the general food distribution in Fino division were very frustrated by the fact that they were mainly involved in the arduous tasks of policing the distribution rather than acting as logisticians.

The MSF team were too focused on the process of controlling the delivery of food and therefore paid insufficient attention to monitoring what happened with the food at household level. This made it difficult to fully assess the impact of the programme, e.g. what food ended up in the household.

We believed from the start that it was very important to involve all the actors in the area (local authorities, government, the community, UN agencies and other NGOs) in the programme. If resources are limited (as they were in this case), then it is even more important that local government contribute resources within their capacity.

^*"indicators show unusual fluctuations outside expected seasonal ranges, and asset levels of households are too low to provide an adequate subsistence level and vulnerability to food insecurity is still high".